Exploring the Psychometric Properties of the Family Empowerment Scale Among Latinx Parents of Children with Disabilities: An Exploratory Structural Equation Modeling Analysis

Abstract

This study examined the psychometric properties of the Family Empowerment Scale (FES) among Latinx parents of children with intellectual and developmental disabilities (IDDs), a population historically underrepresented in empowerment research. Given the cultural and contextual factors that may shape empowerment experiences, Exploratory Structural Equation Modeling (ESEM) was utilized to assess the scale’s structural validity. ESEM supports a four-factor model that aligns with, but also refines, the original structure of the FES. The lack of loading for several items indicates the need for revisions that better reflect the lived experiences of Latinx parents. ESEM provided a more nuanced view of the scale’s dimensional structure, reinforcing the value of culturally informed psychometric evaluation. These results underscore the importance of validating empowerment measures within diverse populations to inform equitable family-centered practices.

1. Introduction

Empowering families is essential to the success of family-centered approaches in healthcare, mental health, and education. It involves supporting families in building the knowledge, skills, and confidence necessary to make informed choices, advocate effectively, and actively participate in services that impact their lives. Empowered families are more equipped to work collaboratively with professionals, navigate complex systems, and contribute to better outcomes for their children and loved ones. As such, promoting family empowerment is recognized as a foundational component of inclusive, equitable, and responsive care systems [1]. Importantly, empowerment is not experienced uniformly across all families. Differences in culture, language, socioeconomic status, and access to resources shape how families define and enact empowerment. For example, Latinx families may experience empowerment differently due to cultural values such as familismo and respeto, community dynamics, and systemic barriers, underscoring the need to draw on the culturally specific literature to better understand and support their empowerment processes. Similar considerations have been raised for other groups, including immigrant families, low-income caregivers, and families from racially and linguistically diverse backgrounds, where empowerment may manifest in distinct ways or face unique challenges [1].

Given the critical role of measuring family empowerment effectively, there is a strong demand for tools that are both reliable and valid across various populations and service environments. The Family Empowerment Scale (FES), initially created by Koren et al. (1992) [2] to assess empowerment among families of children with emotional and behavioral issues, is one of the most commonly used instruments. Over time, the FES has been adapted for use with a broad range of caregivers, including parents of young children in general [3,4], families raising children with disabilities [5], children dealing with chronic illnesses [6] and caregivers of adults with mental health conditions [7].

Prior studies have often been characterized by samples with a disproportionate representation of highly educated, non-Hispanic White caregivers [5,8,9,10]. This raises important questions about the generalizability of empowerment measures, as empowerment may differ across cultural, linguistic, and educational contexts. What is considered an empowering practice in one community may not resonate in another, and certain domains of empowerment (e.g., advocacy, decision-making, and community involvement) may hold different levels of importance depending on cultural values and lived experiences. Given that the construct of empowerment may vary across cultural, linguistic, and educational backgrounds, this calls into question how widely applicable the Family Empowerment Scale (FES) truly is and emphasizes the necessity of establishing the psychometric validity of FES among more diverse caregiver populations. Previous research indicates that comparable instruments, such as the Family Quality of Life scale, may demonstrate reduced effectiveness among socially disadvantaged groups [11]. Therefore, evaluating the psychometric properties and core components of the Family Empowerment Scale (FES) within diverse groups is essential to confirm its relevance and accuracy across different cultural and socioeconomic contexts. This includes families affected by systemic inequities related to race, ethnicity, language, and disability. Enhancing the cultural validity of such instruments is key to promoting more equitable support and services for historically marginalized communities [12].

Examining the factor structure of the Family Empowerment Scale (FES) specifically for Latinx parents of children with disabilities is essential to ensure that the scale is psychometrically valid and culturally appropriate for this population. The FES was originally developed and validated in predominantly White, middle-class samples (e.g., Gerkensmeyer et al., 2008 [8]; Koren et al., 1992 [2]; Huscroft-D’Angelo, 2018 [9]; Lambert et al., 2020 [10]), and its factor structure may not fully capture how empowerment is experienced, expressed, or prioritized by Latinx families, especially those navigating language barriers, systemic inequities, and cultural values such as familismo and respeto.

Without validating the factor structure for this group, researchers and practitioners risk using a tool that may overlook or misrepresent key aspects of empowerment that are culturally and contextually relevant. Confirming or adapting the factor structure helps ensure that the scale accurately reflects how empowerment manifests among Latinx caregivers, identifies the domains most in need of support, and guides the development of culturally responsive interventions and policies. In short, this kind of validation strengthens both measurement precision and equity in service delivery, especially in special education and disability-related fields.

Within this framework, it is crucial to investigate the fundamental aspects of family empowerment among Latinx parents of children with intellectual and developmental disabilities (IDD). To meet this objective, the current study evaluates the psychometric characteristics of the Family Empowerment Scale (FES; [2]) by applying an Exploratory Structural Equation Modeling (ESEM) approach, with a specific focus on a U.S. sample of Latinx parents caring for children with IDD.

1.1. Confirmatory Factor Analysis and Exploratory Structural Equation Modeling

Confirmatory Factor Analysis (CFA) is a widely used method within Structural Equation Modeling (SEM) for evaluating theoretical models that describe the relationships between observed variables and underlying latent constructs. CFA is guided by theory and requires researchers to predefine both the number of latent factors and which observed indicators are associated with each factor. Typically, CFA models are structured so that each indicator loads only on its designated factor, with all cross-loadings constrained to zero [13,14].

Exploratory Structural Equation Modeling (ESEM) is a statistical technique that integrates the flexibility of exploratory factor analysis (EFA) with the confirmatory power of structural equation modeling (SEM). It allows for more realistic representations of latent constructs by permitting cross-loadings (i.e., items loading on multiple factors), which are often constrained to zero in traditional confirmatory factor analysis (CFA) [15,16]. This approach is particularly appropriate for constructs that are theoretically related but not perfectly distinct, as it captures the complexity and overlap often present in social-behavioral data. ESEM enables testing latent structure with greater flexibility than CFA, modeling cross-loadings while improving model fit statistics, and simultaneously estimating measurement and structural relationships [15,17].

1.2. Previous Research on the Family Empowerment Scale

Family empowerment plays a vital role in delivering effective family-centered care across healthcare, mental health, and educational settings. The Family Empowerment Scale (FES), created by Koren, DeChillo, and Friesen (1992) [2], is a widely recognized instrument used to evaluate the empowerment of families raising children with developmental, emotional, or behavioral challenges. The FES is designed to assess how capable families feel in advocating for their child, navigating services, and influencing their child’s growth and surroundings. Comprising 34 items, the scale measures empowerment across three key areas: the Family domain, which reflects confidence and control within the household; the Service System domain, which focuses on relationships with service providers; and the Community/Political domain, which assesses engagement in civic and policy-related matters [2].

The original three-factor structure of the Family Empowerment Scale, introduced by Koren et al. (1992) [2], has been replicated in a few studies employing both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) (e.g., [3,4,8]). However, findings from CFA have been somewhat inconsistent. While some studies have supported alternative models such as a four-factor solution [5] or a three-factor bifactor model, others have highlighted issues such as item redundancy and factor overlap [9,10,18].

A bifactor model aligns well with the theoretical foundation of the FES, which conceptualizes empowerment as a two-dimensional construct—capturing both the degree of empowerment and the manner in which it is expressed [2]. This modeling approach is particularly appropriate in contexts where factors are highly interrelated, as is the case with the Family, Services, and Community subscales. Expanding on these findings is essential to further clarify the FES’s underlying conceptual framework and to confirm its structure through confirmatory factor analysis [12].

Latinx families frequently encounter distinct systemic challenges, including language barriers, immigration-related concerns, and culturally specific norms regarding advocacy and caregiving. These contextual influences may impact how empowerment is perceived and demonstrated, highlighting the need to examine the cross-cultural validity of the FES.

Although the FES is widely used, its psychometric properties have been insufficiently explored in diverse populations—especially among Latinx caregivers—through the use of modern analytical techniques like Confirmatory Factor Analysis (CFA) and Exploratory Structural Equation Modeling (ESEM), marking a significant gap in the existing literature—especially considering ESEM’s capacity to uncover culturally nuanced patterns within psychological constructs. ESEM offers a promising alternative by empirically exploring latent structures by allowing cross-loadings. ESEM can identify poorly performing items or those that do not fit well in the expected factor, guiding scale refinement while preserving psychometric integrity [15,16].

Utilizing ESEM to assess the FES enables a more detailed, realistic, and culturally responsive approach to validation, making it particularly valuable for examining the scale’s structure among Latinx parents of children with disabilities. Conducting such analyses with diverse and historically underrepresented groups, like Latinx caregivers, can provide more accurate and culturally meaningful understandings of empowerment. This type of research is crucial for enhancing the validity and fairness of assessment tools used in family-centered interventions, especially for populations that have traditionally faced marginalization due to disability, language, and ethnic background.

1.3. Purpose of This Study

To the best of our knowledge, no previous research has employed ESEM to investigate the factor structure of family empowerment specifically among Latinx parents of children with disabilities—a group that continues to be underrepresented in the literature. In this study, we applied Exploratory Structural Equation Modeling to the Family Empowerment Scale (FES; [2]) to assess its psychometric performance (internal consistency and structural validity) within a sample of 96 Latinx parents of children with intellectual and developmental disabilities (IDD) in the United States. Additionally, we aimed to identify items that perform poorly or exhibit weak alignment with their intended factors, with the goal of informing potential scale revisions while maintaining psychometric rigor. This analysis is intended to identify the most suitable factor structure for this sample.

2. Method

2.1. Sample

The study received approval from the Institutional Review Board (IRB) before data collection began. Researchers used purposeful sampling to recruit 96 Latinx parents of children with intellectual and developmental disabilities (IDD) from two states in the U.S. Participants were eligible if they self-identified as Latinx—defined as individuals born in or with ancestry from Latin America—,were parents or legal guardians of a school-aged child with a disability receiving special education services, and were enrolled in an advocacy training program. Those who did not identify as Latinx or were not part of the program were not included in the study. The research specifically targeted families with children aged five to eighteen who lived at home, allowing for a focused analysis of parent–child dynamics relevant to the study’s goals. Power analysis was performed using the semTools 0.5-7 package in R [19]. We evaluated statistical power using the population Root Mean Square Error of Approximation (RMSEA), a commonly used fit index in structural equation modeling. Power analysis based on RMSEA assesses the ability to detect poor model fit (i.e., whether RMSEA exceeds a threshold such as 0.05 or 0.08) given the model’s degrees of freedom and sample size. This approach allows for determining whether the study is sufficiently powered to reject an inadequate model, rather than focusing on individual parameter estimates. The results indicate that sample sizes of 96 yielded statistical power estimates of 0.995.

The sample was predominantly female, with 97% (n = 93) of participants identifying as women. The average age of the parents was 40.85 years (SD = 6.85). Most of their children were male (77.1%, n = 74), and just over half (58.33%, n = 56) were 9 years old or younger. The mean age of the children was 9.56 years (SD = 4.72). When asked about their child’s disability, 43.8% (n = 42) of parents indicated a diagnosis of autism. In terms of household income, a significant portion of families (85%, n = 82) reported earning less than $49,000 annually. Regarding education levels, 28.1% (n = 27) of parents had not completed high school, 31% (n = 30) had a high school diploma, and 22.9% (n = 22) had some college experience. Only 17.7% (n = 17) held a bachelor’s or advanced degree.

2.2. Measure

The Family Empowerment Scale (FES), created by Koren and colleagues (1992) [2], is designed to assess empowerment in three primary domains: within the family unit, in interactions with service systems, and in the broader community. The Family Subscale measures the degree of parental engagement in their child’s everyday activities. The Services Subscale evaluates parents’ confidence and perceived influence over services related to their child’s disability. The Community Subscale examines how involved parents are in community-oriented efforts. The scale uses a 5-point Likert response format, ranging from 1 (Not at all true) to 5 (Very true), with higher scores indicating greater empowerment. Scores are determined by summing the responses to individual items. All scales were translated into Spanish by two native Spanish speakers. We conducted the forward/back translation method [20] to ensure the quality and accuracy of each translation. A translated version of FES was given to participants in Spanish. Canino et al. (2008) [21] demonstrated that the FES has strong reliability among Spanish-speaking parents of children with intellectual and developmental disabilities, with Cronbach’s alpha values of 0.93 for the Family domain, 0.89 for Services, and 0.85 for Community.

2.3. Data Analysis

Data analyses were conducted using Mplus 8.10 [22]. There was no missing data. We conducted CFA to assess the original three-factor and three-bifactor models and ESEM with weighted least squares estimation with means and variances adjusted (WLSMV) to investigate and identify the optimal factor structure underpinning the theoretical framework of FES.

Model fit was assessed using the Comparative Fit Index (CFI; [23]), Tucker–Lewis Index (TLI; [24]), and Root Mean Square Error of Approximation (RMSEA; [25,26]). CFI and TLI values above 0.90 are generally regarded as indicating acceptable model fit, while values exceeding 0.95 are considered to reflect excellent fit (e.g., [25,26,27,28,29]). For RMSEA, scores of 0.08 or below are typically viewed as indicating adequate fit. Values at or below 0.05 [25,26] or 0.06 [27] suggest excellent model fit, with a value of 0 reflecting a perfect fit.

Factor loadings represent the numerical strength and direction of the association between an observed variable and an underlying latent construct in factor analysis. They indicate how strongly each item is linked to a factor. Loadings above 0.70 are typically viewed as strong indicators of a factor, values between 0.40 and 0.69 are considered moderate to acceptable, and values below 0.30 suggest a weak relationship with the factor [13].

3. Results

3.1. Descriptive Statistics

Table 1. Descriptive statistics.

All Parents	Mean: Scale	Mean: Item	SD: Scale	SD: Item	Alpha	Omega	Range: Scale
Overall Empowerment	114.32	3.36	24.86	0.73	0.951	0.962	34–170
Family	41.72	3.48	9.65	0.80	0.903	0.906	12–60
Service	41.88	3.49	9.52	0.79	0.894	0.899	12–60
Community	30.73	2.56	7.91	0.66	0.833	0.840	10–50

displays the descriptive statistics and reliability measures for empowerment scores from a sample of 96 participants.

Table 2. Description of FES Items.

FES	Description of Items
Empower 1	Parents’ Right to Approve Child Services
Empower 2	Handling child issues effectively
Empower 3	Feeling capable of enhancing children’s services in my community
Empower 4	Confidence in assisting in a child’s growth and development.
Empower 5	Being aware of the Steps for Concerned Child Services
Empower 6	Ensuring professionals understand my views on my child’s service needs.
Empower 7	Being aware of what to do when problems arise.
Empower 8	Contacting legislators about children’s issues.
Empower 9	Feeling that family life is under control.
Empower 10	Understanding how the children’s service system is organized.
Empower 11	Making good decisions about what services my child needs.
Empower 12	Working with professionals to decide on services.
Empower 13	Staying in regular contact with service providers.
Empower 14	Having ideas about the ideal service system for children.
Empower 15	Helping other families access needed services.
Empower 16	Seeking information to better understand my child.
Empower 17	Belief that parents can influence services.
Empower 18	Valuing my opinion equally with professionals in service decisions.
Empower 19	Expressing my opinions about services being provided.
Empower 20	Providing feedback to agencies and government on improving services.
Empower 21	Confidence in solving problems with my child.
Empower 22	Knowing how to get agency administrators or legislators to listen.
Empower 23	Knowing what services my child needs.
Empower 24	Awareness of parent and child rights under special education law.
Empower 25	Belief that my parenting experience can improve services.
Empower 26	Ability to ask for help when needed with family problems.
Empower 27	Learning new ways to help my child grow and develop.
Empower 28	Taking initiative to find services for my child and family.
Empower 29	Focusing on my child’s strengths as well as problems.
Empower 30	Understanding the service system my child is involved in.
Empower 31	Taking action when problems arise involving my child.
Empower 32	Belief that professionals should ask my input on services.
Empower 33	Understanding my child’s disorders.
Empower 34	Feeling that I am a good parent.

presents a description of the items in the Family Empowerment Scale (FES). Empowerment was evaluated in three areas—Family, Service, and Community—as well as through a composite Overall Empowerment score. The table includes mean scores and standard deviations at both the full-scale and per-item levels, along with reliability estimates using Cronbach’s alpha and McDonald’s omega for each domain.

Overall, participants had an average total empowerment score of 114.32 (SD = 24.86) at the scale level, equivalent to a mean item score of 3.36 (SD = 0.73). Scores for the Family and Service subdomains were comparable, with means of 41.72 (SD = 9.65) and 41.88 (SD = 9.52), respectively. In contrast, the Community subscale yielded a lower mean score of 30.73 (SD = 7.91) and 2.56 (SD = 7.91) at the scale and item levels. All scales demonstrated strong internal reliability, with alpha values ranging from 0.833 for Community to 0.951 for Total Empowerment, and omega values showing slightly higher strength.

3.2. CFA Models

To determine whether a unidimensional structure would be suitable for our data, we first examined a one-factor model. The CFA one-factor model did not fit well (CFI = 0.851, TLI = 0.841, RMSEA = 0.118, SRMR = 0.108), showing that our sample’s score variation could not be well explained by a unidimensional structure. See

Table 3. Summary of Fit Statistics for CFA and ESEM Models.

CFA	CFI	TLI	RMSEA	SRMR
One Factor	0.851	0.841	0.118	0.108
Three Factors	0.863	0.854	0.113	0.105
Three Bifactors	0.929	0.919	0.084	0.069
ESEM
Two Factors	0.925	0.915	0.086	0.07
Two Bifactors	0.934	0.919	0.084	0.06
Three Factors	0.934	0.919	0.084	0.06
Three Bifactors	0.951	0.936	0.075	0.051
Four Factors	0.951	0.936	0.075	0.051
Four Bifactors	0.963	0.948	0.068	0.046
Five Factors	0.963	0.948	0.068	0.046
Five Bifactors	0.97	0.954	0.063	0.04
Six Bifactors	0.97	0.954	0.063	0.04

Note. CFA: Confirmatory Factor Analysis; ESEM: Exploratory Structural Equation Modeling; CFI: Comparative Fit Index; TLI: Tucker–Lewis Index; RMSEA: Root Mean Square Error of Approximation; SRMR: Standardized Root Mean Square Residual.

.

We then evaluated the original three-factor CFA model proposed by Koren et al. (1992) [2], which produced a slightly better fit than the one-factor model (CFI = 0.863, TLI = 0.854, RMSEA = 0.113, SRMR = 0.105). However, the original three-factor model did not meet acceptable fit thresholds (CFI and TLI > 0.90; RMSEA < 0.08), indicating that it does not adequately capture the factor structure of the FES in our current sample of Latinx parents. Then, we used the original subscale-based framework to fit our data to the three-bifactor model. A detailed examination of the factor loadings showed that the family and service domains had uniformly low and negative loadings, suggesting a potential inverted interpretation or lack of coherence, despite the fact that the model fit much improved (CFI = 0.929, TLI = 0.919, RMSEA = 0.084, SRMR = 0.069), as presented in Table 3 and

Table 4. Mean Factor Loadings for CFA and ESEM Models.

CFA Models	Factors	Items	Average Factor Loadings
3 bifactors with original three subscales	G	34	0.64
	Family	12	−0.14
	Service	12	−0.09
	Community	12	0.34
ESEM Bifactor Models	Factors	Items	Average Factor Loadings
2 bifactors	G	30 items; 4–7, 9–14, 15–34	0.69
	F1	4 items; 5, 7, 10, 11	0.5
	F2	8 items; 8, 14, 15, 20, 22, 24, 26	0.55
3 bifactors	G	30 items; 4–7, 9–34	0.68
	F1	7 items; 8, 10, 14, 15, 20, 22, 24	0.58
	F2	3 items; 5, 7, 11	0.49
	F3	2 items; 12, 14	0.45
4 bifactors	G	33 items; 1–7, 9–34	0.64
	F1	2 items; 5–7	0.5
	F2	8 items; 5, 8, 10, 14, 15, 20, 22, 26	0.56
	F3	2 items; 12, 14	0.45
	F4	1 item; 27	0.41
5 bifactors	G	29 items; 4–7, 10–34	0.69
	F1	3 items; 5, 7, 11	0.48
	F2	6 items; 8, 10, 15, 20, 22, 24	0.6
	F3	No items	0
	F4	1 item; 28	0.42
	F5	1 item; 26	−0.45
ESEM Factor Models	Factors	Items	Average Factor Loadings
2 factors (34 items)	F1	12 items; 5, 8, 10, 11, 14, 15, 20, 22–24, 26, 30	0.66
	F2	22 items; 1, 4, 6–8, 12, 13, 16–19, 21, 23, 25, 27–34	0.6
3 factors (36 items)	F1	9 items; 2, 5–7, 9–12, 14	0.6
	F2	17 items; 1, 4, 13, 16–19, 21, 23, 25, 27–29, 31–34	0.6
	F3	10 items; 8, 10, 14, 15, 20, 22–26	0.61
4 factors (35 items)	F1	8 items; 2, 4–7, 9–11	0.58
	F2	8 items; 12–14, 16–19, 21	0.62
	F3	9 items 8, 10, 14, 15, 20, 22, 24–26	0.6
	F4	10 items; 1, 4, 23, 27–29, 31–34	0.53
5 factors (41 items)	F1	7 items; 2, 5–7, 9–11	0.58
	F2	9 items; 11–14, 16–19, 21	0.53
	F3	9 items; 8, 10, 14, 15, 20, 22–24, 26	0.57
	F4	7 items; 1, 3, 4, 21, 24, 32, 34	0.52
	F5	9 items; 16, 17, 25–29, 31, 33	0.55
6 factors (33 items)	F1	7 items; 2, 5–7, 9–11	0.55
	F2	6 items; 11–14, 18–19	0.58
	F3	5 items; 8, 10, 15, 20, 22	0.63
	F4	7 items; 16, 17, 27–29, 31, 33	0.57
	F5	4 items; 23–26	0.61
	F6	4 items; 4, 21, 32, 34	0.55

Note. G: general factor; F1: factor 1; F2: factor 2; F3; factor 3; F4: factor 4.

. These findings imply that, given the characteristics of our sample, the original FES structure proposed by Koren et al. (1992) [2] is not sufficient as a theoretical framework for understanding Latinx parents’ FES scores.

3.3. ESEM Models

Since the fit of the original three-factor model was not acceptable and the factor loadings of the original three-bifactor model were low and negative, we performed ESEM analyses to determine the optimal structure for representing the FES for our Latinx parent sample. All the ESEM models demonstrated acceptable fit to the data, outperforming both the original CFA three-factor subscale-based model and the unidimensional model, as displayed in Table 3. Among the factor models tested, the five-bifactor and six-factor models showed the best fit (CFI = 0.970, TLI = 0.954, RMSEA = 0.063, and SRMR = 0.040) because adding more factors enhanced the model fit, as displayed in Figure 1 and Figure 2. However, an examination of factor loadings indicated several problems, suggesting that incorporating a general factor into the model would not be appropriate. As presented in Table 4, across all the bifactor models, many items loaded primarily onto the general factor but failed to load onto any specific group factors. For example, only four items loaded onto the first group factor and eight onto the second in the two-bifactor model. In the three-factor bifactor model, 22 out of 34 items did not load onto any group factor. Similarly, in the four-factor model, 21 out of 34 items lacked meaningful loadings on group factors. Notably, in the five-bifactor model, no items loaded on the third group factor. Bifactor solutions were not further regarded as suitable structures because of the poor structure and limited interpretability of the group factors as well as the presence of a dominant general factor.

Although the five- and six-factor models yielded excellent model fits, several items in the five-factor model exhibited noteworthy cross-loadings above 0.30, which is in line with the multidimensional nature of empowerment. In the five-factor model, multiple items—including 10, 11, 14, 16, 17, and 24—showed loadings on more than one factor. In comparison, the six-factor model showed cross-loadings for only two items (10 and 11). Some items (e.g., items 4, 8, 10, and 14) appeared in multiple factors (factors 1, 3, and 4) in the four-factor model.

Item 30 did not load significantly on any of the factors (loadings < 0.20), suggesting limited relevance or poor alignment with the latent constructs in this sample. Three items, including 1, 3, and 30 in the six-factor model, did not load significantly on any of the factors (loadings < 0.20). Interestingly, item 3 failed to load onto two-factor, three-factor, four-factor, and six-factor models. Item 30 did not load onto three-, four-, five-, and six-factor models.

As a result, based on statistical model fit, factor loadings, theoretical coherence, and model parsimony, the three-factor ESEM model was selected as the best representation of the FES for our Latinx parent sample. The model fit the data well (CFI = 0.934, TLI = 0.919), and factor loading analysis indicated that it was highly aligned with the latent constructs in this sample. Standardized factor loadings in the three-factor model ranged from 0.41 to 0.88 on their primary factors. Certain items (e.g., items 10, 14, 23, 25) loaded onto multiple factors, which is typical in ESEM where cross-loadings are allowed, acknowledging that some items contribute meaningfully to more than one latent dimension.

Table 5. Correlations for Factors for ESEM 3-Factor Model.

	F1	F2	F3
F1	-
F2	0.45 ***	-
F3	0.403 ***	0.248 ***	-

Note. *** p < 0.001.

presents the correlations among three factors, which ranged from 0.248 to 0.450. The correlation between F1 and F2 was moderate (r = 0.450, p < 0.001) while the relationship between F2 and F3 was small in magnitude (r = 0.248, p < 0.001). These inter-factor correlations suggest that the three domains are related and may reflect an overarching general construct.

In our final three-factor ESEM solution, as presented in

Table 6. Factor Loadings for ESEM 3-Factor Model.

	Items	Primary Factor Loadings
Factor 1: Managing the Child’s Needs and Family Functioning	2	0.51
	5	0.71
	6	0.61
	7	0.79
	9	0.43
	11	0.81
	12	0.55
Factor 2: Parenting Confidence and Service Navigation	1	0.55
	4	0.41
	13	0.54
	16	0.57
	17	0.67
	18	0.72
	19	0.55
	21	0.62
	27	0.83
	28	0.77
	29	0.69
	31	0.60
	32	0.47
	33	0.65
	34	0.67
Factor 3: System-Level Advocacy and Collective Action Beyond the Individual Family	8	0.70
	10	0.54
	14	0.49
	15	0.55
	20	0.82
	22	0.88
	23	0.45
	24	0.65
	25	0.45
	26	0.58

, we assigned items with cross-loadings to the factor on which they had the highest loading, provided the loading exceeded 0.30 and aligned with theoretical expectations. Cross-loadings were interpreted as evidence of multidimensionality rather than item misfit, consistent with best practices in ESEM [16,30].

As displayed in Table 6, Factor 1 consists of 7 items (2, 5–7, 9–12), which likely represent family functioning or personal efficacy—the sense that one can influence decisions and manage challenges effectively within the family context. Factor 2, with 15 items, such as 1, 4, 13, 16–19, 21, 27–29, 31–34, relates to engagement with systems—such as schools or service providers—reflecting a parent’s confidence and ability to advocate for their child within external systems. Factor 3, comprising 10 items (8, 10, 14, 15, 20, 22, 23, 24–26), reflects a blend of advocacy and interaction with broader systems, emphasizing collaboration, influence, and communication with professionals. This involves community or policy-level empowerment, focusing on broader participation or leadership beyond immediate family or school systems.

Two items, 3 and 30, did not load significantly on any of the factors (loadings < 0.20), suggesting they may have limited relevance or do not align well with the underlying FES constructs measured in this sample. These items do not show a strong enough association (loading) with any of the identified factors in the model. Specifically, its factor loadings are too low (e.g., below a threshold like 0.40), suggesting that it does not measure the same underlying constructs as the rest of the items. It may be ambiguous, poorly worded, or not relevant to the population being studied. It might be interpreted differently by participants, especially in culturally diverse or linguistically adapted samples.

4. Discussion and Implications

To evaluate the psychometric structure of the Family Empowerment Scale (FES) among Latinx parents of children with disabilities, this study employed Exploratory Structural Equation Modeling (ESEM) techniques. ESEM offers a unique advantage by integrating the flexibility of exploratory factor analysis (EFA) with the model-testing capabilities of confirmatory factor analysis (CFA) [15,16]. Unlike traditional CFA, which constrains item cross-loadings to zero, ESEM allows items to load on multiple factors [15,16], providing a more realistic representation of complex constructs such as family empowerment. This is especially important in culturally diverse populations, where constructs may manifest differently than in the scale’s original validation samples. Empowerment is not a uniform experience across families; rather, it is shaped by cultural background, language, socioeconomic conditions, and available resources. For instance, Latinx families may engage with empowerment differently, influenced by cultural values such as familismo and respeto, as well as community contexts and systemic barriers. This highlights the importance of drawing on culturally specific scholarship to better capture and support these processes. Comparable concerns have been noted for other populations, including immigrant families, low-income caregivers, and racially and linguistically diverse households, where empowerment may take distinct forms or encounter particular obstacles.

ESEM also yields improved model fit indices and more accurate factor correlations, enhancing construct validity and reducing bias [15,17]. By applying ESEM, this study was able to assess the dimensional structure of the FES in a statistically rigorous yet flexible way, ensuring cultural relevance and structural validity in this underrepresented Latinx parent population.

4.1. Psychometric Properties of FES Scores in Latinx Parents of Children with IDD

The current study extended evidence of the psychometric quality of FES scores to Latinx parents of children with IDD contexts in the following manner. First, our findings indicate that FES scores have strong internal consistency when administered to Latinx parents. In previous validation studies focused on FES (e.g., [2,5,8,9,31]), alpha reliability estimates ranged from 0.85 to 0.94. Patterns of reliability estimates for the current sample were, generally, similar on average to those reported in the previous studies, ranging from 0.83 to 0.95.

Second, the absence of a strong general factor in the bifactor models in our sample may reflect important theoretical considerations regarding the nature of empowerment among Latinx families of children with disabilities. Rather than functioning as a single, unified construct, empowerment in this context may be more culturally and contextually specific, with families experiencing distinct forms of empowerment across domains such as family functioning, community engagement, and interaction with service systems. This finding aligns with the idea that empowerment is not a global trait but a multidimensional and situationally influenced process, particularly for populations navigating cultural, linguistic, and systemic barriers. As such, the results underscore the importance of examining empowerment through a lens that recognizes its variability across environments and experiences.

Third, the results of this study indicated that the ESEM three-factor model may effectively capture the conceptual underpinnings of the FES structure for our Latinx parent sample. Previous factor analysis research for the FES supports the three-factor structure of the FES with no model fits reported (e.g., [3,8]) or a CFI of 0.931 [9]. In our sample, the original three-factor CFA model did not produce acceptable fit indices (CFI = 0.863, TLI = 0.854, RMSEA = 0.113, SRMR = 0.105). These results indicate that the original three-factor structure of the FES does not adequately capture the experiences of Latinx parents of children with IDD.

Instead, family empowerment within this group is more likely to be represented by the three-factor model (CFI = 0.934; TLI = 0.919; RMSEA = 0.084; SRMR = 0.06) identified through ESEM. This revised structure may reflect dimensions such as family functioning or personal efficacy, parental confidence, and service navigation, a combined factor encompassing advocacy and broader system interaction, and empowerment at the community or policy level. Each factor captures a distinct yet interrelated dimension of empowerment, moving beyond the original three-factor model proposed by Koren et al. (1992) [2], as explained below.

4.2. Three-Factor Model of FES for Latinx Parents of Children with IDD

Figure 3 presents an overview of three-factor ESEM model.

Factor 1: Managing the Child’s Needs and Family Functioning. This factor includes 7 items (2, 5, 6, 7, 9, 11, and 12) and appears to center on parents’ perceptions of their effectiveness within the family unit. It reflects their sense of personal agency in managing challenges and making decisions that directly affect their child and household. This dimension underscores the internal or familial component of empowerment, such as confidence in one’s parenting abilities and emotional resilience.

Factor 2: Parenting Confidence and Service Navigation. Comprising 15 items (1, 4, 13, 16–19, 21, 27–29, 31–34), this factor highlights parents’ capacity to engage with external systems—such as schools, service agencies, or healthcare providers. It reflects empowerment through advocacy, communication, and navigation of institutional structures. This is particularly relevant for Latinx caregivers who may face cultural or systemic barriers when accessing and influencing services.

Factor 3: System-Level Advocacy and Collective Action Beyond the Individual Family. This 9-item factor (8, 10, 14, 15, 20, 22, 24–26) reflects a hybrid of advocacy and collaboration with broader systems beyond the individual family. It reflects broader civic involvement, including participation in community activities, leadership roles, and efforts to influence policy or advocate at the societal level. This dimension is critical for understanding how empowerment extends into collective spaces, particularly for communities historically marginalized by systems of care and governance.

4.3. Proposed Revisions to FES Item Wording and Structure

An interesting pattern from ESEM findings was that items 3 and 30 did not align with the three-factor model as well as the four- and six-factor models. Specifically, item 3, “I feel I can have a part in improving services for children in my community”—did not load significantly on any factor in the two-, three-, four-, and six-factor ESEM solutions. This item reflects community-level empowerment, specifically beliefs about influencing systems or services beyond the individual family. If it failed to load on any factor, it may suggest that: first, the construct it reflects is perceived as distinct from the factors identified in our sample (e.g., family-level or service-system empowerment); second, Latinx parents in our sample may not perceive themselves as having influence at the community or systems level—possibly due to systemic barriers, cultural values around advocacy, or limited opportunities for participation; third, the item may feel abstract or aspirational, particularly if parents are more focused on immediate needs (e.g., caregiving, navigating services for their own child); Fourth, terms like “improving services” or “having a part” may be interpreted differently depending on linguistic or cultural factors. This result underscores the importance of considering sociocultural context when interpreting empowerment constructs, particularly in underrepresented populations.

Furthermore, item 30, “I have a good understanding of the service system that my child is involved in” did not load significantly on any factor. Although conceptually aligned with the service system empowerment domain, the item might overlap too much or too little with other items intended to measure service system empowerment. This result may also reflect variability in how participants interpret what it means to “understand the system.” The notion of “understanding the system” might not resonate culturally, especially if trust in institutions is low or if engagement with systems is transactional or mediated through advocates. Some participants may not feel confident about understanding the system at all, especially in complex or fragmented service systems (e.g., multiple providers, language barriers). Latinx parents—especially those facing linguistic, cultural, or systemic obstacles—might not feel equipped to assess their own level of understanding, resulting in low variability or inconsistent responses. Respondents may have interpreted it differently (e.g., “good understanding” might mean emotional, procedural, or legal understanding—leading to inconsistent responses). For some Latinx parents, systemic complexity, linguistic barriers, or lack of formal orientation to services may limit perceived knowledge or confidence, even if they are actively involved. Alternatively, the item may tap a distinct subdimension—such as informational empowerment—that was not well captured by the broader factor structure in this sample. This finding highlights the need to consider how empowerment is experienced and expressed across diverse cultural and linguistic contexts.

The consistent failure of items 3 and 30 to load across multiple factor solutions suggests that these items may not be contributing meaningfully to the internal structure of the scale. This points to potential psychometric limitations, such as weak construct alignment or redundancy. If further analyses (e.g., item-total correlations or reliability diagnostics) confirm their limited value, these items should be considered for revision or removal to enhance the scale’s internal consistency and factorial clarity.

Additionally, in the context of cross-cultural research, non-loading items may reflect discrepancies in cultural relevance, norms, or lived experience. Such findings underscore the importance of evaluating not just statistical fit, but also cultural validity when adapting instruments for diverse populations. Future research should include qualitative studies to better understand why certain items do not resonate with Latinx parents. In addition, longitudinal investigations are needed to examine the longer-term effects of empowerment on service utilization, child outcomes, and advocacy. Future revisions of the scale should carefully examine whether these items resonate meaningfully across cultural groups and modify or replace them as needed.

4.4. Suggested Practices for Factor Analysis

We offer the following recommendations for the application of CFA and ESEM. The research-related flowchart in Figure 4 illustrates the sequence of steps in conducting an Exploratory Structural Equation Modeling (ESEM) study. It begins with identifying the research problem, reviewing literature, and formulating hypotheses, then moves through study design, data collection, and preliminary analyses. The process continues with Confirmatory Factor Analysis (CFA). When CFA models do not demonstrate adequate fit (i.e., CFI below 0.90 or RMSEA above 0.08), the next step is to specify and estimate an ESEM model, assess its fit, interpret the findings, and ultimately report the results and implications.

5. Limitations

This study recognizes several limitations that may affect how the findings are interpreted. It is important to consider these constraints in order to accurately frame the results and inform future research directions. One key limitation is the relatively small sample size, which restricts the generalizability of the outcomes. Structural Equation Modeling (SEM) typically requires a minimum sample size of 100 to 150 participants to yield reliable results and the range of 200–500 to produce significant power [32,33,34]. While smaller samples—particularly those under 100—may still provide meaningful insights in studies involving underrepresented populations, such findings should be interpreted with caution, given the limitations in statistical inference and the restricted generalizability of the results.

Another limitation of the study is the disproportionate number of women in the sample, likely reflecting traditional caregiving roles more commonly taken on by women—particularly within Latinx families of children with IDD. Because mothers are often more directly involved in caregiving, they may possess greater insight into their children’s needs, which can be viewed as a strength. However, this gender imbalance may also limit the generalizability of the findings by underrepresenting the experiences of male caregivers, thereby potentially offering a skewed view of family empowerment. As such, the findings warrant careful interpretation.

A further limitation of this study is that we did not test for measurement invariance across key demographic groups (e.g., gender, language, or cultural background). Without establishing invariance, it remains unclear whether the factor structure of the Family Empowerment Scale functions equivalently across subgroups, which limits the generalizability and comparability of the findings.

To mitigate this limitation, future research should aim to deliberately recruit more male participants to achieve a more balanced sample. Furthermore, it would be beneficial for subsequent studies to replicate the analytic methods employed here using larger and more diverse populations and conduct measurement invariance to determine the extent to which these findings apply across different groups and measurement instruments.

6. Conclusions

This study shed important light on the dearth of research and contributed to the evidence of the FES’ psychometric properties while making unique proposals for an ESEM three-factor structure based on the FES theoretical foundation. This study also identified psychometrically weak items and suggested the revision of the FES for Latinx parents of children with IDD. Using both CFA and ESEM techniques substantially advanced the evidence supporting the psychometric robustness of FES scores and offered valuable direction for future studies examining other multidimensional psychological measures.

Identifying these three dimensions of FES enhances our understanding of the complex ways Latinx parents perceive and express empowerment while raising a child with disabilities. These findings offer useful insights for developing culturally responsive interventions, resources, and policies that better reflect and support the empowerment experiences of Latinx families. Future studies should aim to replicate these results using larger and more varied samples, intentionally include more male caregivers, and examine how empowerment influences service utilization, family well-being, and child development over time. This study helps fill a critical gap in the literature by centering Latinx family perspectives and may lay the preliminary groundwork for building more equitable systems within special education and disability services.